The Dollar and Kraay (2000) paper has proved to be remarkably influential with many of its conclusions widely quoted, particularly in support of the open market policies of the ‘Washington consensus’. However, although there have been a number of critical commentaries there have been very few formal analyses of the results or of the robustness of the support which they provide for the policy conclusions. In this paper the Dollar and Kraay results are investigated from a number of different perspectives. First, a number of questions are raised about the approach adopted. In particular, the Dollar and Kraay paper is notable for having no theoretical structure supporting the specification of the equations. It is unclear how much significance therefore can be attached to the correlations uncovered. In addition, there are the well-known difficulties of drawing conclusions from large cross section samples as well as the attendant problems of data quality. Finally, the identification of poverty with the income of the lowest quintile does not map into either an absolute or relative measure of poverty. There are thus grounds for an initial scepticism. However, this paper then considers in some detail the precise results reported in Dollar and Kraay. The results are replicated and a number of experiments with different regressors and different samples are performed. It is found that the central result of a strong correlation between average per capita income and the income of the lowest quintile is robust and holds under all of the various regressions. However, a number of important caveats are noted. First, a similarly strong result is also found for the higher quintiles. One is entitled to wonder whether the regressions are picking up any movement in the distribution of income, which is known to have changed markedly in a number of countries. Second, the significance of the other regressors in Dollar and Kraay, upon which much of the policy support hinges, changes dramatically under different samples and equations. Although the negative impact of inflation is maintained in most, but not all, of the alternative experiments, the significance of the openness variable vanishes while the significance of the rule of law variable, for which Dollar and Kraay found no evidence, emerges strongly. In addition, when the Gini coefficient is substituted for the income of the bottom quintile the performance of the equation falls markedly, with, however, a strong negative correlation with average income suggesting that higher income reduces inequality. It is unclear how this result is consistent with the Dollar and Kraay findings. The implications of this paper are that in general the policy prescriptions associated with the Dollar and Kraay regressions cannot be sustained. In addition, the weakness of the variable chosen to measure poverty and the differing support provided in different specifications for the other regressors fully justifies the initial scepticism and invites further research in this area.
In order to set up a list of libraries that you have access to,
you must first login
or sign up.
Then set up a personal list of libraries from your profile page by
clicking on your user name at the top right of any screen.